Where the Values are
Localizing Epistemic and Contextual Values in Social Science researchMaster Wijsbegeerte van een bepaald wetenschapsgebied
Emma van Veenen 10645993 August 2019
Supervisor: Dr. Federica Russo Second reader: Dr. Michiel Leezenberg
2
Abstract
In this master’s thesis on philosophy of the social sciences, I explore epistemic values and contextual values as well as their influence on choices of methodology, theory, and, potentially, research results. This thesis was motivated by the current agreement among most scientists and philosophers of science that science is not value-free (Dupré 2007, Douglas 2014, Weber 1949, Longino 1990, Putnam 2002). That science cannot be value free is perceived as a serious problem for social sciences since their objects of research – humans – are not stable or objective. Therefore, the research questions that guide this thesis are as follows: “where are values in contemporary social science?” The associated sub-question is, “what do these values do?” This thesis builds on the extensive literature on the value-ladenness of science. Thereby, the first chapter discusses the role of various types of values in social science research. Then, the second chapter investigates measurement in the social sciences. Finally, the third chapter introduces the case of crony capitalism to argue that values can be localized in contemporary social science research.
3
Table of content
Abstract ... 2
Introduction ... 4
1. Scientific problems with values... 6
1.1 Epistemic values ... 6
1.2 Moral, political, and social values ... 9
1.3 Values in research: inductive risk ... 14
2. Quantification as a search for objectivity ... 18
2.1 Objective science ... 18
2.1.1 Scientific objectivity ... 18
2.1.2 “Trust in numbers” ... 20
2.2 Measurement in the social sciences ... 23
2.2.1 The basic building blocks of research ... 23
2.2.2 Qualitative and quantitative sciences ... 25
2.2.3 Social science and the quest for objectivity ... 27
2.3 Validities ... 29
3. Values in contemporary research: the case of crony capitalism ... 33
3.1 Empirical material ... 33
3.2 Crony capitalism ... 34
3.3 Analysis of empirical material... 36
3.3.1 Influence on the object of research ... 37
3.3.2 Influence on the methodology ... 40
3.3.3 Influence on the conclusions that derive from the evidence... 42
3.4 Discussion of the results ... 45
Conclusion ... 49
Bibliography ... 51
4
Introduction
The goal of all scientists is to deliver interesting and important research that yields worthwhile knowledge. Most social science research affects the world through its policy implications. Reliable policy recommendations require good science. There are myriad definitions of “good” science, but one of the most common is that such science ensures objectivity or, at minimum, is as objective as possible. However, it is also necessary to define objectivity.
Today, scientists and philosophers largely agree that science is not value-free (Dupré 2007, Douglas 2014, Weber 1949, Longino 1990 and Putnam 2002). That science cannot be value-free is perceived as especially problematic for social sciences, as their objects of research, which include democracy and human beings, are often socially informed and thus not stable or objective. To explore this problem, this thesis focuses on values in contemporary social science research. It assumes that if value-free science is an ideal that cannot be achieved, then values exist. Consequently, scientists must learn how to handle these values. To introduce a management technique to engage with different types of values, it is necessary to determine the site and function of those values. Therefore, the research question is, “where are values in contemporary social science research?” The associated sub-question is, “what do these values do?”
This thesis specifically researches the epistemic and contextual values that scientists themselves add to their research. Accordingly, it does not consider other ways in which values are instilled in scientific research. To my knowledge, few researchers have attempted to identify and locate the values of researchers or the influences of such values. Thus, in contrast to the theoretical literature on the value-ladenness of science, the literature on this topic is scarce. In this thesis, I investigate these values to identify and subsequently localize them in social science research.
The first chapter examines the scientific problems with values. To this end, it discusses a variety of epistemic, moral, political, and social values. Furthermore, it introduces an example of a contemporary debate, which concerns the problem of inductive risk, to consider how to address values in social science research. Then, the second chapter, which considers quantification in the social sciences, provides a historical overview to illustrate why the social sciences have employed quantification for objectivity. It also discusses validities that help scientists manage epistemic values. The third and final chapter presents a case study on crony capitalism. This concept was the topic of my master’s thesis on political economy at the University of Amsterdam. In that text, I mostly focused on the definition of the concept and different conceptions of crony capitalism. For my thesis on philosophy of a specific discipline
5 (filosofie van een bepaald wetenschapsgebied), I combined my two research interests, namely political economy and philosophy of science. The case study in the third chapter of this thesis uses the same articles to test whether values can be localized in contemporary research. It concludes that epistemic values can be located by applying criteria for valid measurement and methodological choices, while contextual values (e.g. moral, political and social values) can be located in topic choice. However, the other locations and effects of moral, political and social values are harder to determine. The aim of this thesis is to clarify our understanding of the meaning of value-laden science to indicate how values can be localized and subsequently controlled by scientists.
6
1. Scientific problems with values
This chapter offers a literature review concerning the ways in which scientists account for the value-laden nature of research. First, it introduces two types of values, namely epistemic values and contextual values. The former are discussed in the first section, the latter in the second. According to Longino (1990), contextual values is an umbrella term for moral, political, and social values. It further encompasses the most heavily discussed values in this thesis: moral, political and social values. The third section presents a concise literature review on the problem of inductive risk to exemplify how scientists engage with values in their daily practice.
As a general definition, values refer to “the normative or emotive commitments people hold” (Douglas 2014: 163). They can be tacit or explicit. Thus, the values that people maintain have an impact on how they view the world and what they see. So, for people who hold a different set of values and a contrasting philosophical position, the world is different. Generally, there is an acknowledged difference between epistemic values and political, moral, and social values. Section 1.1 addresses the topic of the former.
1.1 Epistemic values
This section explores types of epistemic values as well as their differences and similarities. Epistemic values determine and influence the kind of scientific knowledge that we can gain. For instance, research that consists of multiple three-hour interviews with 20 subjects over 10 years produces different knowledge than a research that investigates a sample of 10,000 subjects through the use of a survey. Therefore, methodological choices – and, therefore, values – affect the sort of knowledge that one derives from research.
The choice of methodology is not the only factor that influences the type of scientific knowledge that one can gain. In fact, topic choice, available literature, scientific worldview, and philosophical presuppositions all affect the knowledge that scientists can acquire. Presuppositions are pieces of information that are consciously and/or unconsciously presupposed. Thus, they are assumptions that are not explicit, but which influence thoughts and actions and can sometimes be subtly evident. In this thesis, one goal is to identify presuppositions that influence decision-making. For example, if a scientist is a philosophical realist who performs mostly model-based research and believes that all possible knowledge of the world can be attained by observing the world, there is arguably a correlation between her philosophical position and her choice of methodology.
So, one debate in the philosophy of science has concerned whether certain epistemic values can be considered more important, better, or more useful than other values. To consider
7 some of these epistemic values, I incorporate those that have been discussed by Thomas Kuhn (1977: 321-323): accuracy, internal consistency, external consistency, simplicity, breadth of scope, and fruitfulness.
Accuracy: The results should follow other empirical and experimental evidence.
Longino (1996:42) has equated this concept to empirical adequacy.
Internal consistency: There are no internal contradictions.
External consistency: There is consistency with other currently accepted theories.
This concept is sometimes called conservatism (Longino 1996: 43).
Breadth of scope: A theory should be broader than individual observations. It
should be possible to formulate inferences that extend beyond earlier observations or sub-theories. Other researchers have employed the similar notions of explanatory power or generality (Longino 1996: 43).
Simplicity: The research should impart order to the observations by compiling
individual and separate observations into a coherent set.
Fruitfulness – The investigation should offer a contribution in the form of, for
example, a new theory, hypothesis, or problem.
Other authors have introduced their own sets of values. For instance, Longino (1996) has identified the feminist epistemic values of novelty, ontological heterogeneity, mutuality of interaction, and applicability to human needs. From the labels of these values, it is immediately clear that they do not fall completely under the epistemic value category but can also be considered moral, political, or social values.
This multiplicity of epistemic value sets directly highlights the disagreement within the philosophy of science with regard to whether the same epistemic values are of sufficient importance to make their “list.” Both Kuhn and Longino have stated that more values could be included on their “list.” Thus, there is an ongoing discussion of whether and which values lead to reliable, trustworthy research or scientific knowledge at all. This inquiry is further addressed in Chapter 2.
With regard to epistemic values, Helen Longino has explained,
Scientific practice is governed by norms and values generated from an understanding of the goals of scientific inquiry. If we take the goal of scientific activity to be the production of explanations of the natural world, then these
8 governing values and constraints are generated from an understanding of what counts as a good explanation, for example, the satisfaction of such criteria as truth, accuracy, simplicity, predictability, and breadth. (Longino 1990: 4)
Longino (1990:4) has classified such values, which stem from the goal of achieving “good,” reliable science, as constitutive values. In a later work (Longino 1996), she has discussed the difference between cognitive and non-cognitive values, the distinction between which derives from Larry Laudan (1984). In this regard, she has indicated that cognitive and non-cognitive values and constitutive and contextual (her term for moral, social and political values) can be applied interchangeably (Longino 1996: 41), though she has preferred the former since cognitive and non-cognitive values allow for more vagueness (idem: n3). Nevertheless, she has not explained that her preference of a vague boundary between the notions fit with her assumption of no strict distinction between cognitive and non-cognitive (contextual) values because vagueness is more favorable to this grey area.
The decision and approach to regulate presuppositions and epistemic values must be determined by the group of scientists themselves. By critically discussing concepts, assumptions, method strategies, and other topics among themselves, they can diminish the influence of their individual preferences (Longino 1990). A condition for such a peer group is the inclusion of researchers with different views. Otherwise, the researchers may not be able to ask critical questions, as they all assume the same perspective. Furthermore, they should discuss not only methods, topics, and theoretical assumptions but also all parts of scientific research (idem). According to Longino, the intersubjective interaction between researchers is key. However, not every type of interaction suffices. She has specifically identified four features that should be included (idem: 76-79).
1) Recognized avenues for criticism: Institutionalized places for discussion and critique, such as peer reviews. These activities should be considered as important as other parts of the research.
2) Shared standards: To agree and disagree with each other in regard to new methods or rediscovered theories, a peer group should have some shared standards, such as epistemic and social values and substantive principles.
3) Community response: The critical discussion should eventually lead to actual change.
9 4) Equality of intellectual authority: This criterion derives from Jurgen Habermas and holds that no single set of assumptions or presuppositions can be dictated because their proponents have more political or social power than others.
Thus, from Longino’s perspective, scientific knowledge is based on community, as it is only possible to act in accordance with these criteria if one is part of a community of scientific researchers (Longino 1990: 69).
Thomas Kuhn has demonstrated another approach to values that assumes that epistemic values are the only guide by which scientists can determine truth. This assumption follows from the assertion that one cannot “run a test to see how often choosing the more coherent, simpler, and so on, theory turns out to be true without presupposing these very standards of justified
empirical belief (Kuhn 2002: 32).” Kuhn has further claimed that if one considers a set of
epistemic values to be superior to others for producing more reliable and accurate research, then those epistemic values are the exact guidelines by which one can make that choice. In other words, they provide the lens through which one judges “better” or “worse” science (idem: 33).
This section has discussed various epistemic values deriving from Thomas Kuhn and Helen Longino to illustrate that epistemic values are key factors for consideration, and the privileging of one set of epistemic values over others is a value-laden choice. These insights are invoked in the analysis of the crony capitalism case in Chapter 3. The next section discusses the other type of values, which are contextual (i.e. moral, political, and social) values, to apply Longino’s term.
1.2 Moral, political, and social values
This section introduces the so-called opponent of epistemic values, namely contextual (i.e. moral, political, and social) values. This section first defines these values before explaining how scientists evaluate them. Political values are, in the most concrete sense, “the perceptions of a desirable order, and determining whether a political situation or a political event is experienced as favorable or unfavorable, good or bad” (Halman 2007). They render political judgment possible. In the last two decades, political scientists have become increasingly interested in political values and their behavior (idem). Scientists presumably have tacit or explicit ideas about the ideal political order. These ideas can inform their worldview and, accordingly, their perspective of whether political developments are desirable or undesirable. For example,
10 numerous political scientists have researched solutions to resolve seemingly contradictory perceptions of political systems such as democratic Islam or Islamic democracy to allow more countries to become democracies (Bukay 2007, Khan 2006).
Like political values, moral values shape one’s perceptions of right and wrong. There are diverse sets of moral theories that describe which theory corresponds with which type of values as well as which value can be considered superior to the other. Some examples of moral values are as follows:
Care: to protect, nourish and watch over someone, mostly children. Fairness: equal and proportional distribution of love, hours, money, etc.
Loyalty: toward one’s teammates, family, country, or another group, including by
defending this group and caring for the group itself as well as its members.
Respect for authority: hierarchical relations, such as in a family, community, or
nation, that expire when the authority does not act responsibly; not the same as power.
Freedom: to act as one wants and be free from control by others (Haidt 2012:
146-162)
Numerous moral values beyond these five can be identified, and many more theories describe what these values could or should entail.
The last of the contextual values are social values. Like political values, social values reflect the specific perspectives and commitments that people maintain toward social issues. For example, for a long period in history, domestic work was not considered proper work. This perception reflected a social value, namely that work was performed mostly by men and not by housewives. Consequently, gross domestic product (GDP) does not account for domestic work, in part because men exclusively developed the concept of GDP. Only once feminist scientists mentioned the exclusion of domestic work from GDP did it become a point of discussion (Giannelli et al. 2012, Greenstein 2000). Such presuppositions or social values usually become clear only once, in this case, women start working in economics and notice that the activities that they do view as work are not included. This realization has prompted discussions of which factors should be part of GDP and which other types of non- “nine-to-five” work actually constitute work.
Philosophers have disagreed about the degree of difference between contextual values and epistemic values. In summary, epistemic values enable researchers to distinguish between
11 types of knowledge; hence, they can reveal what and how we can know about the world. Meanwhile, political, moral, and social values shape what we think the world is. For most philosophers, only the latter is a possible threat, or values are not even perceived as an existing problem. In his book The Collapse of the Fact-Value Dichotomy and Other Essays (2002), Hilary Putnam has strikingly summarized,
But the history of the philosophy of science in the last half-century has largely been a history of attempts-some of which would be amusing if the suspicion of the very idea of justifying a value judgment that underlies them were not so serious in its implications to evade this issue. Apparently any fantasy-the fantasy of doing science using only deductive logic (Popper), the fantasy of vindicating induction deductively (Reichenbach), the fantasy of reducing science to a simple sampling algorithm (Carnap), the fantasy of selecting theories given a mysteriously available set of "true observation conditionals," or, alternatively, "settling for psychology" (both Quine) is regarded as preferable to rethinking the whole dogma (the last dogma of empiricism?) that facts are objective and values are subjective and "never the twain shall meet. (Putnam 2002: 145)
Although Putnam has been harsh toward empiricists, he advances a valid point. For a long time, scientists and philosophers have supported the dichotomy of facts and values. Putnam was one of the first scholars to argue the contrary. Some have formed such counterarguments from a pragmatist perspective (Putnam), while others have adopted a feminist perspective (Longino, Douglas). Still others have taken on a “new” realism/naturalism perspective (Ladyman and Ross 2009) that assumes that institutions will control for naturally existing values.
Heather Douglas (2014) has made a distinction between legitimate and illegitimate functions of values in research. A legitimate function of moral, political, and social values is that they offer the necessary guidelines and restrictions to research and how it is conducted (Douglas 2014: 169). Thus, ethical considerations are important for research on humans and animals. Currently, methodological courses teach examples wherein ethical and moral values were not respected.
One of the most famous instances, namely the Stanford prison experiment, was adapted into a movie, which we were required to watch for my methodology class at the University of Amsterdam. In this experiment, the social psychologist Philip Zimbardo divided the participants into two groups – prisoners and guards – and placed them in a prison that was constructed on the Stanford University campus (Zimbardo 1973: 243). However, the experiment was terminated after six days, as the student volunteers fulfilled their assigned roles
12 so seriously that the prisoners were significantly (mentally) harmed, and the guards became overly harsh and cruel (Haney et al. 1973). Although the results offer key insights into the importance of social context for behavior, this experiment (and the later-discussed Milgram experiment) prompted a serious discussion of the ethical boundaries of research and, according to Stephen Reicher and Alexander Haslam (2006), led to a decline in similar studies (Douglas 2014: 168).
Nevertheless, there are illegitimate roles of contextual values as well. Moral, political, and social values can illegitimately influence the methodological choices of researchers by permitting moral values and presuppositions to guide the research. This creates biases that compromises the epistemic values that are also involved (Douglas 2014). Moreover, values can indirectly influence research through the necessary uncertainty that is inherent to all scientific research. For instance, regardless of the size of the research population, there is always a chance that the sample is skewed. This possibility reflects the problem of inductive risk, which is further discussed in the next section.
As the preceding section has noted, Longino (1996) has suggested that the epistemic values that Kuhn has introduced can be replaced or supplemented by feminist values. According to Longino, when employed within science, feminist values promote an equal balance in power relations between genders. Although such values are not necessarily specifically feminist, they serve “feminist cognitive goals.” Thus, Longino has offered an approach to changing one’s institutional surroundings (1996: 50). Furthermore, she has argued that a comparison of the values of Kuhn with her feminist suggestions reveals that both “import” political values (idem: 54). Still, she has clarified that
in specific research contexts the traditional [Kuhn’s] virtues have a demonstrable political valence. I don’t want to say that the traditional virtues are always politically regressive, but that the fact that they sometimes are means that we cannot treat them as
value-neutral grounds of judgment. [Emphasis added]
This crucial point is foundational to this thesis. Not all research is value-neutral; in fact, most (if not all) research is value-laden. Such a starting assumption dictates that no scientist can treat science as though it is value-neutral. Therefore, researchers are obliged to explore and develop an understanding of values as well as their specific role in scientific research.
Putnam (2002) has asserted that facts and values should not be viewed as two totally separate worlds. Rather, the above-described epistemic values should indeed be considered
13 values (Putnam 2002: 31). The sharp distinction between values and facts originates from Hume and was further advanced by logical positivists (idem: 27), the latter of whom “did not regard ethics as a possible subject of rational discussion” (idem: 29). Later, in 1939, they reconsidered the role of ethics and values in research. In 1950, Quine's abandonment of the notion of factual and conventional statements left the fact-value dichotomy was in ruins (idem: 30).
Putnam tried to resolve the chaotic situation by arguing that epistemic values are values as well, and that they are a precondition of scientific inquiry itself. However, not only epistemic values inform the way people view the world; according to Putnam, so-called ethical values relate to the same issues (idem: 32). Therefore, he has introduced the notion of “thick” ethical concepts (idem: 34), which highlight the connectedness of facts and epistemic and ethical values. To explain this notion, Putnam has cited the example of the word “cruel” to illustrate that, for most people, it is intuitively clear that this word contains both normative and moral suppositions. Hence, the notion of “cruel” crosses the border of the fact-value dichotomy by sometimes functioning as a description and serving as a moral judgment (idem: 35). Still, Putnam has concluded that this insight has not led to diminished use of the fact-value dichotomy among philosophers; the dichotomy is still practiced – just in different terms.
Following the philosopher Bernard Williams, most philosophers continue to argue that one can describe the world with exclusively scientific terms. Moreover, because such terms are not value-laden, scientific research or results are not value-laden either (Putnam 2002: 40). An argument of Williams, according to Putnam, is that he “does not see how to provide us with a
metaphysical explanation of the possibility of ethical knowledge” (idem: 44). Putnam agrees
with the idea that it is not at all useful to explain how ethical knowledge is possible in absolute terms. However, this does not mean that ethical values do not play a role in science.
The history of the fact-value dichotomy indicates that the discussion still maintains an “absolutely not” or “absolutely yes” framework with regard to the role of values in scientific research. However, a stable metaphysical explanation of the possibility of ethics, moral knowledge, or values is not imperative to conclude that, in the daily practice of science, both epistemic and moral values have a significant impact that can be legitimate or illegitimate. In view of this, one can conclude that science is never value-free. Although the institutional foundation of science does offer firm grounds for eliminating or becoming aware of values, the selection of a set of epistemic values is necessarily a value-laden choice. Hence, ignoring these and contextual values in scientific research can ultimately endanger the institutional foundations of science.
14 This section has addressed multiple types of moral, political, and social values as well as how scientists have perceived them. The discussion has demonstrated the importance of values for scientific research and the obligation of scientists to understand when and how they influence research. Furthermore, there are differences and similarities between epistemic and contextual values. Notably, the distinction between the two types is not that harsh, and both have shaped conceptions of all elements of research practices. In view of these findings, it could be insightful to examine both types of values in the case study on crony capitalism.
1.3 Values in research: inductive risk
The two previous sections have introduced the notions of epistemic and contextual (i.e. moral, social, and political) values. The literature on this specific topic in the philosophy of science has rarely extended beyond the discussion on value-free or value-laden science. However, as argued in the introduction, the notion of value-free science is merely utopic. Thus, only describing the different values and how they are perceived is insufficient to clarify the influences of values in social sciences research. Moreover, it does not offer guidance with which individual scientists or brands of science can navigate values in social science research. Therefore, this section details one useful approach by which philosophers of science can discuss values in practice: the problem of inductive risk.
In the book Exploring Inductive Risk, Case Studies of Values in Science (2017), Douglas writes in the foreword, “Inductive risk is the chance, the pervasive possibility, of getting it wrong in an inductive context (ix).” In science, researchers can employ inductive or deductive reasoning. Most social science research is inductive; thus, it starts with observations and builds upon them to form a theory (though research is always theory-laden in some sense). In contrast, deductive research first identifies the theory before proceeding to individual observations. In the end, most (if not all) research involves an “inductive move” whereby it establishes an inference that extends beyond the object of observation (i.e. the previously discussed epistemic value of breadth of scope) (Elliott and Richards (eds.) 2017: 1). In science, every choice – from the topic and measurements to the assessment of the stability of the data – can be epistemically wrong. Thus, the scientific discussion of inductive risk concerns how to manage such risk. As discussed above, values form the guidelines for which evidence is sufficient to accept a hypothesis. Accordingly, “the risk of getting it wrong – the inductive risk – is what values need to assess” (idem: x). According to Douglas, assessing this inductive risk entails not only epistemic values but also contextual (i.e. moral, political, and social) values.
15 The first researchers to fully define inductive risk were the philosophers Richard Rudner (1953) and Charles West Churchman (1948a, b), though early indications can be found in the work of William James (1896). Churchman has introduced the concept of pragmatic inference to align the theory of statistical inference with the practice of conducting research. A key element of pragmatic inference is that “the scientific method is conceived as an activity designed to choose the most efficient means for one end” (Churchman 1948a: 267). Therefore, pragmatic considerations are connected with statistical measurements. However, in some cases, the goals of the research cannot be attained at the same time, which necessitates the development of a theory to determine which ends should be prioritized. In this regard, Churchman has concluded that a proper theory of induction requires consideration of not only effectiveness and probability (through statistical inference) but also ethics to establish which ends are important (idem).
Rudner (1953) has refined the ideas of Churchman (Elliott and Richards (ed.) 2017: 2) in his influential article entitled “The Scientist qua Scientist makes Value Judgements.”1 This
article can indeed be perceived as the starting point for discussion of inductive risk, though Rudner’s elaboration on inductive risk was actually inspired by a discussion between Carnap and Quine regarding ontological commitment (Rudner 1953: 1). In the article, Rudner focuses on the fact-value dichotomy. Since he does not fully accept the arguments to account for values in science, he presents a different argument. First, he states that one certainty in science is that scientists must accept or reject hypotheses (idem: 2). As discussed earlier and assumed by Rudner, no hypothesis can ever be 100% confirmed; hence, it is necessary to have guidelines to decide which evidence is sufficient to validate a scientific hypothesis (idem). According to Rudner, “how sure we need to be before we accept a hypothesis will depend on how serious a mistake would be” (idem). Thus, this decision is ethical in nature.
One example is the variety of percentages in statistics for determining the probability that a statistical inference is inaccurate. This probability is called the statistically significant rate. The conventional standard in social science research is 0.05%. Consequently, there is a 95% certainty that the relationship that was found in the sample can be generalized to the population (Bryman 2014: 347), and positive results are random (and thus not correct) in five out of 100 cases. In addition, two other standard levels are used: a 99% or 99.9% statistically significant rate. The three rates are differentiated by *, **, and *** in the result tables of social scientific
1This led to some other interesting titles: The Scientist qua Policy Advisor Makes Value Judgements (2012) from Katie Steele or Must the Scientist Make Value Judgements? from his opponent Isaac Levi (1960).
16 research; * signifies at least 95% significance, ** indicates at least 99% significance, and *** denotes at least 99.9% significance.
Although the 95% significance is the standard for the political science department in which I studied, it is not applied by all scientific disciplines. Will Koehrsen (2018) has observed that a significance level (p) of 0.1 to 0.001 is most commonly used at his lab, Cortex Intel. Moreover, the scientists who discovered the Higgs boson particle employed a p-value of 0.0000003 (Koehrsen 2018). With this p-value, the results are completely random in one out of 3.5 million cases. Even in the social sciences, where a statistically significant level of 0.05 is the standard, it is common to explicitly note when the level is 0.01 or 0.001. Therefore, it can be concluded that the standard significant level in the social sciences cannot be considered a neutral standard, as the choice of level of statistical significance is also an epistemic choice. This conclusion further supports that science cannot be considered value-free.
Naturally, Churchman and Rudner also received criticism, with Richard Jeffrey (1956) and Isaac Levi (1960, 1962) as the most influential critics (Elliott and Richards (eds.) 2017: 3). Levi (1960) has focused on so-called “Rudner-Churchman assumptions” in arguing that neither philosopher has proven that a scientist should make value judgments to assess the severity of the possible mistake. His key claim is that “the value-neutrality thesis depends upon whether the canons of scientific inference dictate assignments of minimum probabilities in such a way as to permit no difference in the assignments made by different investigations to the same set of alternative hypothesis” (Levi 1960: 357). Therefore, there are ways to maintain the standards for sufficient evidence. In contrast with Levi, Jeffrey has focused more heavily on Rudner’s claim that the profession of scientists entails accepting and rejecting hypotheses. He has posited that it is not the role of the scientist to make that decision. Rather, it is the responsibility of the scientist to precisely determine the probability and then allow the people who are affected by the decision, namely the public, to make that final choice (Elliott and Richards (eds.) 2017: 3). In his 1965 essay “Science and Human Values,” Carl Hempel combines the work of critics and proponents by asking two questions: Can ethical, moral, or political questions be answered by objective empirical science? And can objective empirical science offer ethical, moral, and political criteria to live by? (Hempel 1965: 82). His answer to the first question is no, but the second question is answered affirmatively. Moreover, Hempel argues that values are influential in scientific inquiry but specifies that this does not necessarily imply that values should always play a role in the results of the research (idem: 93-96). Thus, he argues for a relative judgment of value. If the decision concerns a moral issue, certain values should be involved. However, such sets of judgments are not unchangeable. Instead, they are relative in
17 time and in the degree to which the scientist has to engage with a moral issue (Hempel 1965: 96).
After Hempel’s publication, the discussion temporarily stalled until Heather Douglas re-introduced it to the scientific agenda with her article “Inductive Risk and Values in Science” (2000) (Elliott and Richards (eds.) 2017: 3). As demonstrated above, she has argued that the influence of values extends beyond the hypothesis part of the research. Thus, inductive risk is also relevant to elements of scientific inquiry beyond the hypothesis (idem). Douglas’ work led more authors to become interested in inductive risk and values in science (e.g. Cranor 2008, Wilholt 2009, Elliott 2011, and Biddle 2016 as proponents and Betz 2013 and Lacey 2015 as opponents). One example, “Exploring Inductive Risk” (2017), consists of multiple case studies on inductive risk and values in science. Strikingly, most case studies focus on natural science topics, such as pesticide use, the Higgs boson particle, and climate change.
In future research, it could be interesting to explore not only the nature of inductive risk, the counterarguments against inductive risk, and strategies to handle such risk, as Elliott and Richards have suggested (2017: 261), but also the connection of values in the social sciences with objectivity and measurement within the concept of inductive risk. The latter two are also topics of the next chapter of this thesis.
This subsection has introduced one useful way in which philosophers of science have framed historical discussions of values, objectivity, and methodology by applying a “risk” perspective. This perspective has generated new insight into the role of values in scientific research and its relation to objectivity. This subsection has indicated that this thesis discusses values from a methodology-based perspective. A combination of perspectives could be fruitful further research.
Overall, this chapter has introduced types of values and highlighted the importance of both types for shaping and influencing not only how we think about science but also how we conduct it. The third section of this chapter has addressed inductive risk as one direction for further work in the area of values in (social) science.
18
2. Quantification as a search for objectivity
This chapter explores the question, “if science cannot be value-free, then what constitutes objective science?” To answer this inquiry, the chapter considers the role of objectivity and measurement in science. In regard to objectivity and measurement in science, methodologists and philosophers of science have developed multiple tools and rules to manage and restrict the role of values in social science research.
Section 2.1 on objectivity science presents several perspectives of scientific objectivity. These perspectives reflect differences in philosophical assumptions. Moreover, they uphold disparate views of the definition of values as well as which values should be corrected for and how. Section 2.1.2 then employs the work of Theodore M. Porter in progressing from objective science to a new scientific method: quantification. Subsequently, Section 2.2 addresses measurement in the social sciences. Section 2.2.1 differentiates between concepts, conceptions, and conceptualization, which are the basis of scientific research, Section 2.2.2 briefly discusses the division between qualitative and quantitative science, and Section 2.2.3 answers the question of why the social sciences can be considered a science. Section 2.3 then introduces the notion of validities as the various rules by which scientists can manage values. Currently, validities are vital tools for social scientists to navigate values in their research. These validities are also used to analyze crony capitalism in Chapter 3.
2.1 Objective science
This section first outlines three types of scientific objectivity, each of which reflects a distinct philosophical scientific perspective. The perspectives are interpreted as possible reasons and values that justify social science research. Subsequently, the section progresses from objective science to measurement by introducing the history of quantification through an analysis of the work Trust in Numbers by Theodore M. Porter.
2.1.1 Scientific objectivity
This subsection elaborates on three types of objectivity, namely objectivity as factfulness, objectivity as the absence of normative commitments, and objectivity as freedom from personal biases.
Scientists have not agreed on a conception of scientific objectivity. Broadly speaking, three conceptions of objectivity can be distinguished (Reiss and Sprenger 2017). The first perceives scientific objectivity as factfulness: in the world, there are facts, and it is the responsibility of the scientist to discover them. The second conception of scientific objectivity
19 is that of objectivity as the absence of normative commitments. The third conception accepts that value-free science is not possible but states that science can still be objective if it is free from the personal biases of scientists.
The main proponents of objectivity as factfulness have been Nagel and Williams. The philosophical impetus for this view is the conviction that we, as individuals, differ in our perspectives of the world, but the world does not change. Thus, the world is not dependent on people who perceive it to be “true” or “there.” This view was later adopted by the empiricists such as Popper and Carnap, who have argued that experiments could determine which theory was right and which was not. However, Kuhn has disagreed with this view. According to Kuhn, observations are colored by scientists’ presuppositions. Hence, the observation that confirms the theory is not the only possible one, as it is influenced by what scientists think they see, what they want to see, and what the instruments allow them to see (Reiss and Sprenger 2017: 4-16). The second conception is that of the absence of normative presuppositions of the scientist. It dictates that science is objective as long as moral, social, and/or political values do not influence the research or the scientist. Therefore, it considers epistemic values to be acceptable because they do not influence the objectivity of the research (see Chapter 1 for more details on values). Reiss and Sprenger have stated that the involvement of values is problematic only at the stages of the collection of evidence and the acceptance or dismissal of theories.
The third conception is that science is objective if it is free from personal biases, which refer to the biases of the researchers themselves. This conception does not imply that personal biases do not exist but rather that they should not be included in scientific research by scientists or through social processes (idem: 38). Reiss and Sprenger have presented mechanical objectivity and other approaches to minimize personal biases.
The next subsection details the concept of mechanical objectivity as developed by historian of science Theodore M. Porter. Reiss and Sprenger have utilized the work of historians of science Daston and Galison (1992), who have authored an influential work on mechanical objectivity entitled “Objectivity.” In some respects, the text is consistent with Porter’s description of mechanical objectivity, as they both support a relation between quantification, quantitative measurement, and the removal of personal biases and values by removing the human factor. However, a historical analysis by Porter has concluded that quantification was introduced and pursued to foster trust and does not necessarily lead to fewer values.
The second means of reducing personal biases is the use of social control in science through networks of scientists and institutional codes. Longino’s four features of a well-functioning scientific community (see Chapter 1) can function as an objectivity measure in this
20 regard. The degrees to which these criteria incorporate (1) recognized avenues for criticism, (2) shared standards, (3) community response, and (4) equality of intellectual authority) also determine the extent of objectivity among that research community. Thus, for Longino, scientific knowledge is necessarily community based; therefore, scientific knowledge also provides social knowledge (Longino 1990: 79). Her criterion for objectivity is “a method of inquiry is objective to the degree that it permits transformative criticism” (idem: 76).
2.1.2 “Trust in numbers”
This subsection engages with Porter’s work Trust in Numbers: The Pursuit of Objectivity in
Science and Public Life (1996) to analyze the relationship between the search for objectivity
and the rise of quantification in both science and public institutions. It ultimately argues that upholding rules is a crucial mechanism for fostering trust.
In this book, Porter, a historian of science, explains how numbers attracted our trust and the relation between numbers, trust, science, and policy. Porter wants to answer the question: what is the reason that the social sciences adopted the quantitative methods to a high extent (on which more in Section 2.2)? The most common explanation for the broad adoption of quantitative methods is that all sciences tried to follow the same methodology as that of the natural sciences. This attempt was, and still, is especially strong among much younger social science fields (see Section 2.2.3). For most people – not only scientists – the model of the natural sciences resembles the model of how to conduct proper science.
A logical question follows: why are the natural sciences considered the model of “good” science? Furthermore, which factor(s) caused the rise of quantitative methods in science? Porter has started his investigation from a unique perspective. Specifically, he has researched the relationship between government policy, politics, and trust and changes in scientific research methodology rather than focusing on the rise of quantitative methods. This approach supports an analysis of the increase in popularity of quantitative methods from a broader and not strictly scientific perspective.
According to Porter, the rise of quantification in the methodologies of both the social sciences and the natural sciences is relevant to the need for objectivity. Policymakers, politicians, business leaders, and scientists continuously seek out ways to gain people’s trust. This trust is crucial because it is the foundation of a working (democratic) society. Moreover, a lack of trust from citizens has prompted governors to explore other means of proving that their ideas, policy proposals, and scientific theories are appropriate. Numbers have a seemingly “objective” quality, which can generate such trust.
21 In his book, Porter describes three types of objectivity: absolute, disciplinary, and mechanical. Absolute objectivity is the most common philosophical definition of objectivity and corresponds with objectivity as factfulness, as previously described (Porter 1996: 3). This perspective deems a result, a research “objective” when it describes what is constant in the world – the common structure or object from all perspectives. Meanwhile, disciplinary objectivity refers to “an ability to reach consensus” within a specific community of researchers. Finally, mechanical objectivity has been defined as “following the rules” (idem: 4), which is a definition with which Daston and Galison (1992) can agree.
Of the three types, Porter has chosen to elaborate on mechanical objectivity in his research. Although all types of objectivity are impossible to achieve, he has reasoned that mechanical objectivity is the most interesting form of objectivity because it is relatively the most feasible. Furthermore, mechanical objectivity, which entails adherence to the rules, clarifies the relation between numbers and objectivity. Such adherence accounts for the subjective ideas, presuppositions, and opinions of researchers. Porter has cited judges as one example of mechanical objectivity, as they are only objective to the extent to which they honor the rules of the law (idem).
The process of quantification is a strict process whereby phenomena are translated into numbers, and rules are key to this shift. Hence, “even at their weakest point – the contact between numbers and the world – methods of measurement and counting are often highly rule-bound or officially sanctioned” (idem: 5-6). Thus, the degree of mechanical objectivity is determined by the rule-bound process of quantification. According to Porter, since the introduction of the modern natural sciences in the 17th century, businessmen, politicians, and policymakers have identified this method as capable of acquiring trust. Such use of numbers, statistics, and rule-bound measures has provided support for policy proposals.
That rule is that the measure that makes judges objective also holds for scientific research and explains the emphases of scientific education on methodology and scientific research on quantification. When a scientist follows the prepared methodology, he or she adheres to strict rules; hence, the results of the research are more objective or structured compared to those of studies in which the researcher diverged from the pre-set methodology.
At the same time, quantification obscures a considerable part of the process. Consequently, most quantitative research is not comprehensible for people who are not educated in this field of science. Thus, the rise of numbers was accompanied by an increase in expertise and the emergence of a group of experts. According to Porter, expertise refers to “a relation between professionals – often academic scientists or social scientists – and public
22 officials” (Porter 1996: 6). Therefore, on the one hand, mechanical objectivity provides a measure by which public officials and scientists can “answer to a moral demand for impartiality and fairness [because] [q]uantification is a way of making decisions without seeming to decide” (idem: 8). On the other hand, mechanical objectivity offers public officials and scientists a form of authority that they cannot acquire in other ways (idem), which leads to tension between the mechanical objectivity of quantification and the simultaneous obscuring of such mechanism by that process. Knowledge that is derived through this approach is consequently more difficult to check. This problem of trustworthy knowledge is not only scientific but also – and mainly – moral and social in nature (idem: 11).
Another key aspect of scientific objectivity is scientific trust. Following the globalization of science, a scientific truth is necessarily a collective product (idem: 12). Only knowledge that is validated by another specialist – and preferably by more than one – can be considered scientific knowledge. In We Never Have Been Modern (1993), the philosopher Bruno Latour discusses how global communication has changed the practices of science. An example of such communication over distance is a peer review in which at least two fellow scientists have to agree on at least the design before the results of the research can become an accepted scientific truth. In some way, the process of peer review also relates to Longino’s ideas regarding social scientific knowledge through criterial discussion in a peer group.
The creation of a stronger state and the search for trust have led to the standardization of measures in meters and kilos. According to Porter, this shift was the result of a similar need that prompted the cooperation between scientists and bureaucrats (Porter 1996: 22-23). As a consequence of the standardization, it is possible for citizens to buy and sell goods without the need to check if wheat in a traded box is tamped or loose. As a further development, it is no longer necessary to meet trading partners in person for the transfer of goods. Thus, standardization has fostered trust by making it more difficult to engage in fraudulent behavior. The persistent lack of trust in and among politicians, policymakers, and scientists has stimulated them to develop additional, more complicated methods, rules, and procedures to progress toward the “accounting ideal” (idem: 50).
This subsection has discussed the work of Porter to illustrate the relationship between objectivity, trust, and quantification in the sciences and public bureaucracy. Later, Chapter 3 recalls the relation between objectivity and quantification to analyze connections between methods, validities, and the case of crony capitalism. As Porter has argued, the search for
23 objectivity still leads to quantification in social science research. The next section further develops ideas about quantification and measurement specifically in the social sciences.
2.2 Measurement in the social sciences
The following three subsections elaborate on the multiple aspects of measurement in the social sciences. The first subsection differentiates between concepts, conceptions, and conceptualizations, as these three notions form the basis of all research. The second subsection argues that the widely used distinction between qualitative and quantitative research can be explained by the search for objectivity in the social sciences. However, the distinction should not be accepted as a strict one. The third subsection further details the historical endeavor for legitimate, objective science in the social sciences.
2.2.1 The basic building blocks of research
This subsection elaborates on the building blocks of research: concepts, conceptions and conceptualizations. Scientific measurement entails a progression from the researched concept to a conception of that concept and then to a conceptualization of that conception and ends with an operationalization of that conceptualization.
Measurement is based on the object of research, which is usually a concept. Concepts are core components of our thinking as well as the basis of meaning. Concepts and terms are sometimes mistaken for synonyms but are in fact distinct. Terms are words that express concepts, whereas concepts are “the basic ideas that give sense to a term or expression” (Olsthoorn 2017: 155 emphasis added by Olsthoorn). Hence, if a term does not denote a concept, then the term is meaningless. Concepts classify, categorize, and group the world in such a way that we can understand and speak about it. Furthermore, a non-true concept does not exist, they can be strange, vague or unusable (Olsthoorn 2017). A problem with measurement in research can already emerge at the concept level, as scientists might use the same term to represent different concepts.
The next step involves conceptions, which refer to “interpretations of concepts developed by adding specifying principles and criteria” (idem: 160). Thus, conceptions are structured interpretations of a concept. Like concepts, conceptions cannot be false or true. The various workings of (different) languages are relevant in this context, as conceptions can simply be linguistic differences or different priorities. Therefore, a conception is not a “copy” of reality but rather a device that scientists can apply for a particular purpose (Brown 2014: 4). For example, capitalism is an economic system that is based on private ownership; however, one
24 individual might argue that the state is a necessary condition for the existence of a private ownership system, while another may envision capitalism without the state as a necessary condition for private ownership. Both individuals would believe that their respective conception of capitalism properly describes the meaning of the concept (van Veenen 2018: 3).
Next in the research process is conceptualization. A conceptualization of a conception is the list of attributes that characterize the given conception. Thus, it reflects the process of constructing conceptions. The list of attributes, namely the conceptualization, are transferred into measurable objects, the indicators. This process is the operationalization of the conceptualization.
In the social sciences, concepts can be conceptualized in numerous ways, as the manifestations of such phenomena in the world can be viewed from countless perspectives. As a consequence, there can be multiple and potentially an enormous amount of conceptions for one concept. Furthermore, the existing values and presuppositions of scientists can manifest in this stage of the research. Olsthoorn (2017) and Gerring (2012) have affirmed that social science concepts and conceptions are rarely, if ever, value-neutral. The enormous amount of possible conceptions is one of the reasons why it is important for a scientist to justify the use of one conception over another. Such justification explains why one specific set of indicators provides a more suitable characterization of the concept compared to another set.
Brown (2014) has demonstrated that conceptions can have different goals, which in turn signal unique sets of values that are considered important. His research on varying conceptions of politics has reported a broad range of conceptions – and purposes of those conceptions – in the specific field of politics in science and technology studies. In his article, he introduces three conceptions of politics but begins with five approaches to the study of politics in science and technology studies (STS). The initial focus is on the primary purposes of these five approaches, but it is difficult to determine if the purpose leads to the conception of politics, or vice versa. It stays unclear since two of the five conceptions of politics remain unspecified. A conclusion based on Brown’s work is that different approaches to politics in (STS) vary in the conceptions that they employ (Brown 2014: 9). Brown could probably say a bit more about why that is the case; however, it illustrates the challenge for scientists to develop overviews of utilized conceptions. Furthermore, it underscores the importance of social interaction between scientists who use different approaches. This type of research helps to become and remain aware of the range of conceptions that are applied for the same concept and can preclude misunderstandings. Finally, it can facilitate the location of values in social science research.
25 This subsection has explained concepts, conceptions, and conceptualizations. The differences between these notions are crucial, as they are the basis of scientific research. In the shifts from a concept to a conception and a conception to a conceptualization, values can be influential. The next section explicates how the use of these terms differs between the qualitative and quantitative sciences.
2.2.2 Qualitative and quantitative sciences
This subsection distinguishes between qualitative and quantitative science to argue that such distinction was once useful in the quest for objectivity, but it now obscures the similarities between scientific research approaches. Although many writers have dismissed this distinction as outdated or simply false, social methodologists seem persistent. The following quote derives from the book Social Research Methods by Alan Bryman (2012: 35), which is used in the social science education program of the University of Amsterdam:
However, there is little evidence to suggest that the use of the distinction is abating and even considerable evidence of its continued, even growing, currency. The quantitative/qualitative distinction will be employed a great deal in this book, because it represents a useful means of classifying different methods of social research and because it is a helpful umbrella for a range of issues concerned with the practice of social research. (Bryman 2012: 35)
Thus, the qualitative-quantitative distinction is applied not only to distinguish methods but also to engage with other issues of science. One example can be found directly from Bryman in Table 1:
Quantitative Qualitative
Principal orientation to the role of theory in relation to research
Deductive, testing of theory Inductive, generation of theory
Epistemological orientation Natural science model, in particular positivism
Interpretivism
Ontological orientation Objectivism Constructionism
Table 1: table 2.1 from Bryman (2012): 36
This table indicates that the choice between qualitative and quantitative methods is not only a choice between different types of methodology, according to Bryman’s handbook, which is
26 widely used with social science students, but also directly related to one’s philosophical position. Another discussion of the relationship between quantitative and qualitative methodology and philosophical positions has been presented by Russel Hathaway (1995), who has also offered a summary table (p. 542). It teaches students that the distinction is significant and, for the identity of the social sciences, sufficiently important to teach it to its students. Flipping the table provides information about the philosophical position of a researcher. Then, the applied method reveals insight into the philosophical orientation and interests of the researcher. This is used in the case study in Chapter 3.
Table 2: table 3.1 from Bryman (2012): 76
Bryman (2012) has also created a useful table for the classification of methodology types as qualitative or quantitative. This table is employed for analysis of the case study of crony capitalism in Chapter 3 through comparison with philosophical positions and validities.
Research Design Research Strategy
Quantitative Qualitative
Experimental Typical form. Most researcher
using an experimental design employ quantitative
comparisons between experimental and control groups with regard to the dependent variable.
No typical form.
Cross-sectional Typical form. Survey research
or structured observation on a sample at a single point in time. Content analysis on a sample of documents.
Typical form. Qualitative
interviews or focus groups at a single point in time. Qualitative content analysis of a set of documents relating to a single period.
Longitudinal Typical form. Survey research
on a sample on more than one occasion, as in panel and cohort studies. Content analysis of documents relating to different time periods.
Typical form. Ethnographic
research over a long period, qualitative interviewing on more than one occasion, or qualitative content analysis of documents relating to different time periods.
Case Study Typical form. Survey research
on a single case with a view to revealing features about its nature.
Typical form. The intensive
study, by ethnography or qualitative interviewing, of a single case, which may be an organization, life, family, or community.
Comparative Typical form. Survey research
in which there is a direct comparison between two or more cases, as in cross-cultural research.
Typical form. Ethnographic or
qualitative interview research on two or more cases.
27 As the subsection on Trust in Numbers has indicated, quantification has become increasingly important in the sciences. The social sciences in particular have intensified their focus on the potential reliability of quantitative methods, which could partly explain why the distinction between qualitative and quantitative methods became institutionalized to such a large extent (Montuschi 2014: 135). This distinction also reflects the status that was and is still associated with certain research fields. The scientific statuses of anthropology and economics, for instance, were increasingly driven apart. As a result, original qualitative methods have also been transformed into quantitative formats. For example, some case studies are divided into sections that can be analyzed statistically, and ethnographic rapports can be codified and thereby represented by numbers (idem).
As Porter has demonstrated, the foundation of this distinction does not involve any factual difference between the reliability and objectivity of the methods. Rather, it reflects a quest for trust and different preferences concerning epistemic values. However, most research currently does not fit neatly within these qualitative-quantitative borders. Therefore, the strict distinction that is emphasized in a social science education does not accurately reflect the actual way in which research is performed. Furthermore, the classical distinction between quantitative and qualitative science can offer direction in the search for values.
This subsection has differentiated between qualitative and qualitative science. This distinction is still applied by many researchers and can signal their philosophical position. This insight is used in Chapter 3.
2.2.3 Social science and the quest for objectivity
This subsection further investigates why the social sciences have searched for trust and objectivity. The findings are key to understanding the importance of a “quantitative” set of epistemic values in the social sciences. Furthermore, they support the idea that there is a need for the social sciences to abandon moral, political, and social values in their research.
The precise definition of science and why it is the most reliable methodological approach to acquiring knowledge have fueled debate among philosophers of science for decades. This debate has generated numerous questions: In what way does methodology contribute to objectivity? What difference can methodology make? What can science offer society, both practically and concerning the intrinsic pursuit of knowledge? And how does this relate to ethics? Moreover, researchers have discussed criteria for a distinction and the possible
28 differences between the natural sciences, the social sciences, and the humanities. To this end, they have questioned the influences of such categorizations of science.
As one effect, the relatively new social sciences were driven to prove that they were as reliable as other fields of science. Montuschi has dated the modern natural sciences to the 17th century and the social sciences to the 19th century (Montuschi 2014: 123). In the latter, there was a fear of not being as “scientific” as the natural sciences. The modern natural sciences developed in the 17th century by philosophers such as Francis Bacon, Galileo Galilei, and René Descartes, while the social sciences emerged in the 19th century (idem). By then, the natural sciences had already earned their place in society. Furthermore, the modern natural sciences were based on ideas of “natural” philosophers who believed that, with a proper method, knowledge and values could be separated (idem). Hence, a goal of the newly developed social sciences was to offer factual knowledge about ways to engage with social action without being influenced by (individual) values (idem: 142).
The main difference between the natural and social sciences concerns their topics. The natural sciences study phenomena that are “natural” and thus found in nature. Notably, these sciences are distinct from the exact sciences, such as theoretical mathematics and informatics, in which researchers can engage from an armchair, which applies to most types of philosophy as well. While social science topics can also be found in nature, they are fundamentally different, as they are human, which implies both methodological and ethical effects.
When this “new” type of science was introduced, the initial feedback was that it should copy the method and style of the natural sciences, as they constituted the “proper” way to derive scientific knowledge (Montuschi 2014: 123). Thus, the social sciences had to immediately compete with and compare themselves to the natural sciences and their methods. Although the natural sciences did not suffer in the same way, they did and do experienced their own problems regarding this popular image.
As the social sciences developed, it became clear that they are fundamentally distinct from the natural sciences and thus entail other methods and ethical considerations. The social sciences established a kind of middle ground between the humanities and the natural sciences with the consequence of becoming perceived as less objective compared to the natural sciences (Montuschi 2014). Eleonora Montuschi has identified three points on which the social sciences have focused in their search for objectivity. The first is a need for emphasis on “real” objects and “real” facts. The second is in line with the topic of this thesis, namely, to minimize values in scientific inquiry. The third is to only use methods that are most likely to gain “true” results (Montuschi 2014: 124-125).
